Valeri Alexiev, Jeanna Balreira and Vanessa Moreno Analyzing Structures and Dynamics of Protein/Lipid Interactions Using Voronoi Tessellation Methods Valeri Alexiev, Jeanna Balreira and Vanessa Moreno Numerical Analysis 3351 FINAL PRESENTATION - MAY 3, 2013

Outline Motivation Project Overview Methods Results Further Research References

Motivation The misfolding of proteins and their aggregations has been linked to cause diseases. The surface area, volume, and voids of molecules help characterize the protein misfolding and aggregation on cell membranes. Conformal annular lipids surround a protein, this leads to miscalculations for these parameters using traditional methods

Project Overview In order to better understand why some protein misfolding form harmful plaques... Develop Voronoi tessellation methods to calculate surface area and the volume of different functional groups of lipid and protein molecules. Determine if and how void spaces can be calculated in these lipid and protein groups.

Methods To understand the VT methods used in Voro++ and Voroprot, we: Created our own VT in Matlab Given: A 2D plane setting Generator 1 x1=(.25, .25) ; w1= .25 Generator 2 x2=(.75, .75) ; w2=.50 Calculated the Power VT (Voro++) and Additively Weighted VT (Voroprot).

Methods - 2 generators Power VT for the two generators using our Matlab script Weighted VT for the two generators using our Matlab script

Methods - 'N' generators Points Used: [ (.15, .15), .25 ; (.35,.35),.5; (.5,.5), .3; (1.15,1.15, ), .5; (2.75,2.25) ,.25] Power VT for 'N=5' generators using our Matlab script Weighted VT for 'N=5' generators using our Matlab script

Methods- Checking Accuracy of Algorithms Modified our Matlab code Changed the Power/Additively Weighted distance to Euclidian distance Assigned weight of .25 to all of our points. Points Used: (.25,.25), (.50,.50), (1.75,.75) (1.5,1.5), (2.75,2.5).

Methods- Checking Accuracy of Algorithms Using Matlab's "voronoi" command, we plotted the points' exact Euclidean VT. We can then check our plot with Matlab's "voronoi" plot. Image for Matlab's Voronoi command on points (.25,.25), (.50,.50), (1.75,.75) (1.5,1.5), (2.75,2.5). Both plots on top of each other

Methods- VORO++ Script to Attach Weights Using Python and the VORO++ library, we created a new PDB file with the weights of each molecule attached.

Results- VMD VMD Method a computer visualization program for large biomolecular systems using 3D graphics and built in scripting. Upload the results of weighted VT from Voro ++ Show lipid shells and interfacial water molecules Method Upload the C2_29 protein onto VMD Upload the files POPCHOLSOL_1-9, the lipid shells one by one. We do the same for the interfacial water molecules The 1st solvent shell is adjacent to the protein/lipid, and 2nd solvent shell is the solvent neighboring the 1st lipid shell ... etc.

Results- VMD Output VMD image for the C2_29 protein and its 9 lipid shells VMD image for the C2_29 protein

Results - Frequency of Weights POP Molecules

Results - Means and Atom Distribution POP Molecules

Results - Frequency of Weights CHO Molecules

Results - Means and Atom Distribution CHO Molecules

Results - Frequency of Weights CHO_O6 Molecules

Results - Means and Atom Distribution CHO_06 Molecules

Results - Frequency of Weights SOL (Solvent) Molecules

Results - Means and Atom Distribution SOL Molecules

Results - Voroprot Voroprot "Find buried balls" procedure A software tool used for constructing and visualizing Voronoi diagrams. Can be used to find cavities and pockets in proteins. Can we use Voroprot to find void spaces? "Find buried balls" procedure Investigated the meaning of the parameters of the search. Minimal buried ball radius should be set to 0 as we want to detect even the smaller void spaces. Rolling probe radius determines the size of the probe that explores the cavities.

Results - Voroprot The cavities are not equivalent to void spaces generally, but with the right parameters it should be possible to detect voids.

Further Research - Use weighted PDB for visual analysis How can we visually analyze the nearest neighbor weights in the shells around the protein?

Further Research - Use weighted PDB for visual analysis How can we visually analyze the nearest neighbor weights in the shells around the protein?

Further Research - Use weighted PDB for computational analysis How can we computationally analyze the nearest neighbor weights in the shells around the protein? 'r' code analyze weight frequencies, averages and standard deviations of the 9 shells and their nearest neighbors

Further Research - Use VORO++ to compute radical (power) containers How can we analyze the volume of a container, constructed using VORO++'s power tessellation method? VORO++ library's radical.cc sub-section: x: 1 to 10 y: 1 to 10 z: 1 to 15 number of blocks x = 5 y = 5 z = 5

Further Research - Use Voroprot to find void spaces Use domain-specific knowledge to find the best value for the rolling probe radius. Investigate how including solvent molecules in the pdb file changes how Voroprot detects cavities.